论文部分内容阅读
本文提出的声码器设计方案的特点是:(1)声道参数抽取,采用改进的“协方差—斜格”的线性预测方法。(2)音调周期抽取,采用逆滤波和降低取样率的平均幅差函数方法。(3)参数编码和语言合成,采用声道参数的非均匀最佳编码和增益与音调的对数编码及三角波或伪随机数为激励的声管模型合成。(4)硬件实现,采用位片微处理机和硬件乘法器为主的系统。
The vocoder proposed in this paper is characterized by: (1) channel parameter extraction, using improved “covariance-skew grid” linear prediction method. (2) pitch cycle extraction, the use of inverse filtering and reduce the sampling rate of the average amplitude function method. (3) Parameter coding and language synthesis, using the non-uniform channel coding parameters and coding of the gain and pitch logarithm and triangular wave or pseudo-random number for the excitation of the sound tube model synthesis. (4) hardware implementation, the use of bit-chip microprocessor and hardware-based multiplier system.